Meta-Analytic Validation of the Dissociative Experiences Scale (Des)
Total Page:16
File Type:pdf, Size:1020Kb
Clinical Psychology Review, Vol. 16, No. 5, pp. 365-382, 1996 Copyright © 1996 Eisevier Science Ltd Pergamon Printed in the USA. All rights reserved 0272-7358/96 $15.00 + .00 PII 80272-7358(96)00006-2 THE MEASUREMENT OF DISSOCIATION IN NORMAL AND CLINICAL POPULATIONS: META-ANALYTIC VALIDATION OF THE DISSOCIATIVE EXPERIENCES SCALE (DES) Marinus H. van IJzendoorn and Carlo Schuengel Center for Child and Family Studies, Leiden University, The Netherlands ABSTRACT. The Dissociative Experiences Scale (DES) hos now been used in over 100 studies on dissociation. This article reports on a series of meta-analyses to test some of the theoretical assumptions underlying the DES and to examine the instrument's reliability and validity. Studies wilh the DES were identißed through Psychlit, Medline, Social Sciences Citation Index, and Current Contents. Across sludies in similar domains (e.g., studies on multiple personality disorders) combined effect sizes were computed using the Rosenthal-Mullen approach. The DES showed excellent convergent validity wilh other dissociative experiences questionnaires and inter- view schedules (combined effect size: d = 1.82; N = 5,916). The DES also showed impressive pre- dictive validity, in particular concerning dissociative disorders (Multiple Personality Disorder: combined effect size d = 1.05; N = l, 705) and traumatic experiences (post-traumatic stress dis- order: combined effect size d = 0.75; N = 1,099; and abuse: combined effect size d = 0.52; N = 2,108). However, the discriminant validity was less well established. The DES is sensitive to response and experimenter biases. It is recommended to average DES-scores over more points in time and over morejudges. The DES seems to measure the current view on past dissociative expe- riences. The model of dissociation äs a form of autohypnosis failed to receive support from the data. A developmental model to Interpret dissociation is proposed. DISSOCIATION IS OFTEN considered a psychological defense mechanism for vic- tims of traumatizing events such äs sexual molestation, natural disaster, or combat (Putnam, 1991). It is suggested that it allows these victims to compartmentalize per- ceptions and memories, and to detach themselves from the füll impact of the trauma. At a later stage, dissociation may delay the necessary working through of the traumatic experiences (Spiegel, 1991). In the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; American Psychiatrie Association, 1994) the essential feature of Dissociative Correspondence should be addressed to Professor Marinus van IJzendoorn, Center for Child and Family Studies, Leiden University, P.O. Box 9555, NL-2300 RB Leiden, The Netherlands. 365 366 M. H. van IJzendoorn and C. Schuengel Disorders has been defmed äs a disruption in the usually integrated functions of con- sciousness, memory, identity, and perception of the environment Bernstein and Putnam (1986) developed a self-report measure to assess the degree to which subjects experience dissociation — the Dissociative Experiences Scale (DES). It consists of 28 items, and subjects are asked to indicate on a visual analogue scale which percentage of the time they experience dissociative Symptoms. The scale items were con- structed using data from Interviews meeting DSM-III criteria for dissociative disorders and from consultations with clinical experts in the treatment of dissociative disorders. The DES has been used äs a screening Instrument, distinguishing potential dissociative- disorder patients from other psychiatric patients, and äs a research Instrument. Is it possible to measure individual differences in dissociative experiences reliably and validly by means of a self-report Instrument such äs the DES? This question is central to the current meta-analytic validation of the DES. First, we assessed the instrument's con- vergent validity across studies. Several questionnaires and interview schedules have been developed to measure dissociation, dissociative experiences, or related phenomena. Questionnaires, such äs the Perceptual Alteration Scale (PAS; Sanders, 1986), the Tellegen Absorption Scale (TAS; Teilegen & Atkinson, 1974), the Questionnaire of Experiences of Dissociation (QED; Riley, 1988), the Bliss-scale (Wogan, 1995), the Dissociative Questionnaire (DIS-Q; Vanderlinden, Van Dyck, Vandereycken, Vertommen, & Verkes, 1993), and the dissociation scale of the Trauma Symptom Checklist (TSG40- DIS; Gleaves & Eberenz, 1995), have somewhat different formats and content but their authors all claim to assess some aspect of dissociation or a related phenomenon. Interviews based on the DSM-III-R criteria for dissociative disorders, such äs the Structured Clinical Interview for DSM-III-R Dissociative Disorders (SCID-D; Steinberg, 1994) and the Dissociative Disorders Interview Schedule (DDIS; ROSS, Heber, Norton, Andersen, Anderson, & Barcher, 1989), measure dissociative Symptoms and syndromes that should be associated with DES scores. Second, the DES should show discriminant validity (Campbell, 1960). Subjects from different socio-economic, racial, and cultural backgrounds should perceive the DES items in similar ways if their scores are used for comparison. Dissociation is not expected to be sex-related. One may expect weak asso- ciations with age (Strick & Wilcoxon, 1991). Lastly, the DES should not be affected strongly by response biases: that is, systematic error variance caused by social desirabili- ty or expectancy effects. Third, the reliability of the DES is an important issue. The data on the internal consistency or alpha reliability of the Instrument are summarized, and the studies on test-retest reliability are discussed. Furthermore, the existence of DES subscales are addressed. The proof of the pudding, however, is in assessing the instrument's predictive valid- ity (Nunnally, 1978): Is the DES related to the syndromes for which it has been devel- oped äs a screening Instrument? As Bernstein and Putnam (1986) predicted, the DES should be related to dissociative disorders, such äs Multiple Personality Disorder (MPD), Dissociative Disorder Not Otherwise Specified (DDNOS), and other Dissociative Disorders (ODD), and at the same time it should not be associated with psychiatric disturbances without dissociative characteristics, such äs Personality Disorders, Affective Disorders, Anxiety Disorders, Schizophrenia, or physical disor- ders such äs Seizure Disorder. Furthermore, äs dissociation is conceptualized äs a defense mechanism, it should be present in cases where defense is functional — at least in the short run. Therefore, the DES may be expected to discriminate Post- Traumatic Stress Disorder (PTSD) or Abuse subjects from other (psychiatric or nor- mal) subjects. There has also been speculation on the basis of clinical reports that Eating Disorders may include certain dissociative components (Demitrack, Putnam, Dissociative Experiences Scale 367 Brewerton, & Brandt, 1990), and that subjects who tend to be hypnotizable also tend to have more frequent dissociative experiences (Bliss, 1986). Lastly, because the DES has been applied in several different diagnostic groups, including more than 12,000 subjects, we found it useful to derive some normative data from the extant studies, and to compare the means of those groups. METHOD Data Base Pertinent studies were identified through Psychlit, Medline, and the Social Sciences Citation Index, äs well äs through the "snowball" method (Müllen, 1989). Recent studies were identified using Current Contents. Studies were found through comput- er searches using the keywords "dissociative," "experiences," "scale," "DES," and "dis- sociation," and by using the reference to the Bernstein and Putnam (1986) article. Criteria for inclusion of a study in one of the meta-analyses were äs follows: (a) The DES was used to compare nonclinical or clinical groups; (b) Means or medians were reported; and (c) There was sufficient statistical detail to compute effect sizes. The selection procedure yielded 22 studies on the relation between DES and alter- native measures for dissociation, 34 studies on the discriminant validity of the DES, and 79 studies on the predictive validity of the DES. Only published papers were included; because a separate Journal for studies on dissociation started at around the same time äs the DES became available (Dissociation: Progress in the Dissociative Disorders), probably most DES studies have been published. In 1993, Carlson, Putnam, ROSS, and Torem (1993) published their multicenter study on the validity of the DES äs a screen- ing Instrument for MPD. This study was not included in our meta-analysis, because their sample and the other — separately reported — samples in the meta-analysis would almost certainly overlap. The number of hypothesis tests for the separate meta- analyses varied between 11 and 53, which is sufficient for the computation of com- bined effect sizes (Müllen, 1989; Van IJzendoorn, 1995). The data that were extracted for the meta-analyses from the individual studies are available from the first author. Meta-Analytic Procedures In primary-level studies the unit of analysis is the subject; in a meta-analysis of several primary-level studies the unit of analysis is the outcome of those studies. Because of this difference in unit of analysis, the meta-analytic approach has to be based on a dif- ferent set of statistical techniques. These techniques should, for example, take into account the fact that data in a meta-analysis are usually based on different